Hi @giax! Welcome to the world of SPA! It’s great news that you’re already able to see density for your substrates. I have a few recommendations in addition to what others have already suggested:
If each of your asymmetric units can bind/unbind/catalyze reactions independently, I think your instinct to try symmetry expansion is spot on. This will let you treat each asymmetric unit of the enzyme separately which will likely improve your resolution if some particles have half bound and half unbound conformations.
Once you symmetry expand the particles, it’s important to only perform local refinements. This prevents accidentally creating duplicate particles in your particle stack!
I would try performing symmetry expansion and then making a mask around just one of the asymmetric units (whichever one looks better in your un-expanded maps) and performing a local refinement. This will give you your “consensus” map, where all of your asymmetric units are “stacked on top” of each other. In other words, you’ve lined everything up as best you can while ignoring the fact that some asymmetric units may be bound to substrate and others in the apo state.
From here, I would try performing a 3D classification with the same mask covering just one asymmetric unit. This job skips aligning the particles (since you’ve already done that by this point) and just focuses on classifying them between different maps. Since it doesn’t perform alignments, it can handle a far greater number of classes than something like heterogeneous refinement, which also needs to align the particles.
My first instinct is 3D classification over 3D variability analysis in this case since we’re looking for a binary switch — the enzyme is either in the apo state or the bound state. We know that there are likely continuous conformational changes between those two states, but most particles will be in one or the other. You can, of course, try both and see which works better!
Briefly, I want to mention that when you’re making a mask that cuts through protein density (like you will be here), it is very important that the mask has a soft edge. This prevents “ringing” artifacts which severely degrade the alignment quality. We recommend a minimum soft padding width of 5 * resolution / pixel size, all in angstroms. Since your map is currently going to about 3 angstroms, 15 / pixel size would be a good place to start, but I’d try a few different padding widths and seeing what works best. We have more advice on making masks in our guide!
I hope that’s helpful, and please come back with any questions you still have, or that come up as you run the jobs! I will look in the literature to see if I can find some good examples for you, but hopefully this is enough to get you started!